E1DA Cosmos ADCiso Review

[restorer-john]

A little bit different (but similar question): Can I plug the line out of my APT/Holman preamp into it to digitize my LP's onto my computer?

The input impedance of the Cosmos ADC is rather low:

The Holman is rated to drive 5k (it's output R is ~330R) and above loads, however, it has been tested with 600R loads and apart from a very slight increase in THD, it performed well. Obviously you will lose about 3dB of gain with such a load and the low end is affected with a greater roll-off.

 

[MLGrado]

Upon the hint here at ASR that the Cosmos ADC can be ordered at Aliexpress I looked at the price which is 357 EURO for the A grade version. Due to lack of documentation, only 1 channel with low input impedance and 3.5 mm imput jack to me the price is too high. For double the price QuantAsylum offers 2-ch ADC and DAC including measurement sofware plus support. OK, the low distortion performance is not reaching the cosmos ADC. But for many tests it will be usable.

get the quantasylum and add the E1DA APU. You may have to do the math yourself; adding back the proper amount to compensate for the 1khz notch. But it will let you have all the features you want for most DUT measurements and for those that have more range than the QA, at least you can get extended results with THD+N with the APU

 

[EJ3]

The Cosmos ADC is a general purpose Line in device, so yes.

Thank you!

 

[EJ3]

The input impedance of the Cosmos ADC is rather low:

View attachment 383385

The Holman is rated to drive 5k (it's output R is ~330R) and above loads, however, it has been tested with 600R loads and apart from a very slight increase in THD, it performed well. Obviously you will lose about 3dB of gain with such a load and the low end is affected with a greater roll-off.

Ah, thank you for a bit more information!

 

[MLGrado]

For a good example of how well the entire E1DA chain works for measurements, here is the SMSL SU-1, which measures better than I could ever imagine an approx. 80 US dollar DAC. The combination of the Cosmos Scaler/Buffer, with the APU pre-amp plus notch at either 1khz or 10khz, with the latest ADC with ISO (I run everything off battery power) really does a nice job with REW. (Multitone as well, but REW allows you to compensate for the 1khz notch in software)

 

[Piere]

Received mine from Audiophonics France. It came without any note where to find further information etc. First testing isn't very encouraging and shows following noise performance at SR = 48kHz, both inputs loaded with just 100 ohm resistors (but with or without did not made much difference):

This an overlay of one measurement when the isolated part is powered by a powerbank and the other from the laptop itself. No difference at all. Quite weird responses, not what I expected.

When fed from my Topping E30II DAC with a 1kHz sine at -9dB FS:

So AD performance by itself looks on par. But noise performance rather weird with a strong 1/f content. Do I miss something or just a faulty part?

 

[IVX]

Received mine from Audiophonics France. It came without any note where to find further information etc. First testing isn't very encouraging and shows following noise performance at SR = 48kHz, both inputs loaded with just 100 ohm resistors (but with or without did not made much difference):

View attachment 388002

This an overlay of one measurement when the isolated part is powered by a powerbank and the other from the laptop itself. No difference at all. Quite weird responses, not what I expected.

When fed from my Topping E30II DAC with a 1kHz sine at -9dB FS:

View attachment 388006

So AD performance by itself looks on par. But noise performance rather weird with a strong 1/f content. Do I miss something or just a faulty part?

Please, read the product description here https://e1dashz.wixsite.com/index/cosmos-adc
I guess, your ADC is in the RIAA mode, you can switch that to STEREO or MONO by the Volume slider.

 

[Piere]

Thanks for your reply IVX and for providing that link! Just tested and it happened when volume setting is between 40 an 60. Anything >60 solved the issue. I wasn't aware of a kind of "RIAA Mode". It's working fine now! With REW up to the full 768kHz sampling rate under W10

B.t.w. IMHO any thread about this device on ASR should start with the link given to the product description

 

[IVX]

as you can see, all my posts have the link to our discord channel where tonnes of practical info could be found.

 

[Salt]

as you can see, all my posts have the link to our discord channel where tonnes of practical info could be found.

The signatures are not available on the mobile version, so one has to switch to desktop-website.

 

[bachatero]

Here's the datasheet for the ADC chip used:

https://www.esstech.com/wp-content/uploads/2022/09/ES9822PRO_DS_v0.3.pdf

It says the typical stereo THD+N is -117dB for 48kHz, but the grade B claims 122 - 3 = 119dB for stereo and grade A 123 - 3 = 120. Does this mean that even grade B units are performing better than claimed by ESS?

 

[capslock]

Parts performing better than spec is nothing new. The famous AKM ADC AK5394A was one such part that consistently outperformed its data sheet.

Ivan has written several times that the ESS parts very and some he just has to toss. So in a way, you are seeing only part of the Gaussian distribution.

 

[trl]

Here's the datasheet for the ADC chip used:

https://www.esstech.com/wp-content/uploads/2022/09/ES9822PRO_DS_v0.3.pdf

It says the typical stereo THD+N is -117dB for 48kHz, but the grade B claims 122 - 3 = 119dB for stereo and grade A 123 - 3 = 120. Does this mean that even grade B units are performing better than claimed by ESS?

There's a teardown thread where Ivan was also posting a mod to heat-up the chip to further improve the 3rd harmonic, if really needed.

Basically, providing better power regulation and a very well designed PCB can make the chip to act a bit better. However, like most ICs, there's a Min and a Max for each of the specs, this is why there are more E1DA ADC grades.

 

[bachatero]

Basically, providing better power regulation and a very well designed PCB can make the chip to act a bit better. However, like most ICs, there's a Min and a Max for each of the specs, this is why there are more E1DA ADC grades.

What's weird is that I can't actually find the grade C available anywhere, only grades A and B. I know in the CPU industry that the worse bins sometimes get discontinued because yields improve enough that there are simply no bad chips. Is that's what's happening here?

 

[trl]

In the CPU industry manufacturers are testing all their chips and, even if some chips are raising the bar of...let's say 4GHz operation, they can still sell them as 3.5GHz CPUs because this is what the market is requesting at that specific moment. This is one of the reasons why some of us can easily overclock some of our CPUs. I suppose same applies to ESS chips as well, hence some Cosmos ADCs might have a higher performance than others.

 

[bachatero]

Do you think it might be worth upgrading to a grade A from B? I have a B on order (not shipped yet) and I'm so anxious about missing out on the extra SNR.

 

[CedarX]

In the CPU industry manufacturers are testing all their chips and, even if some chips are raising the bar of...let's say 4GHz operation, they can still sell them as 3.5GHz CPUs because this is what the market is requesting at that specific moment. This is one of the reasons why some of us can easily overclock some of our CPUs. I suppose same applies to ESS chips as well, hence some Cosmos ADCs might have a higher performance than others.

I believe Ivan (@IVX) deserves some of the credits as well! . The OEM "reference schematics" are not always the absolute bests... and, in the case of E!DA products, Ivan goes through quite a bit of iterations to optimize his layouts before releasing them... Others may be happy and stick with "good enough"...

 

[trl]

Of course he does, after all he designed the PCB layout and he is the one choosing the right components and the proper schematic to achieve such a good THD+N results.

 

[Salt]

Do you think it might be worth upgrading to a grade A from B? I have a B on order (not shipped yet) and I'm so anxious about missing out on the extra SNR.

Depends on the purpose.

 

[trl]

Do you think it might be worth upgrading to a grade A from B? I have a B on order (not shipped yet) and I'm so anxious about missing out on the extra SNR.

It really depends on what you want to do with it, so can you please give us an example on how do you want to use it?

Do you own already a Grade B and is not performant enough for your needs? Even so, you might have a look to the E1DA APU as well, because it will increase measurement performance with 30 dB on dedicated frequencies.